Shape optimization problems for variational functionals under geometric constraints

Similar documents
Shape optimization under convexity constraint

Symmetry results for overdetermined problems on convex domains via Brunn Minkowski inequalities

Murat Akman. The Brunn-Minkowski Inequality and a Minkowski Problem for Nonlinear Capacities. March 10

Around the Brunn-Minkowski inequality

Brunn-Minkowski inequality for the 1-Riesz capacity and level set convexity for the 1/2-Laplacian

Brunn-Minkowski inequality for the 1-Riesz capacity and level set convexity for the 1/2-Laplacian

Brunn-Minkowski inequalities for two functionals involving the p-laplace operator

A FABER-KRAHN INEQUALITY FOR THE CHEEGER CONSTANT OF N-GONS

PROD. TYPE: COM ARTICLE IN PRESS. Andrea Colesanti. Received 4 August 2003; accepted 8 June 2004 Communicated by Michael Hopkins

Convex Geometry. Otto-von-Guericke Universität Magdeburg. Applications of the Brascamp-Lieb and Barthe inequalities. Exercise 12.

From the Brunn-Minkowski inequality to a class of Poincaré type inequalities

On the spectrum of the Laplacian

IN SEARCH OF A MAHLER-TYPE INEQUALITY FOR THE EXTERNAL CONFORMAL RADIUS SPUR FINAL PAPER, SUMMER 2015

On the role of energy convexity in the web function approximation

Citation for the original published paper (version of record):

On the spectrum of the Hodge Laplacian and the John ellipsoid

The Isoperimetric Problem in the Heisenberg group H n

Regularity for functionals involving perimeter

SOME REMARKS ABOUT ANALYTIC FUNCTIONS DEFINED ON AN ANNULUS. 1. Introduction. What else? Perimeter, eigenvalues of the Laplacian, etc...

Isoperimetric inequalities and variations on Schwarz s Lemma

DAR S CONJECTURE AND THE LOG-BRUNN-MINKOSKI INEQUALITY

Remarks on Extremization Problems Related To Young s Inequality

Sharp bounds for the bottom of the spectrum of Laplacians (and their associated stability estimates)

A Brunn Minkowski theory for coconvex sets of finite volume

arxiv: v1 [math.ap] 3 Jan 2011

(Non)local phase transitions and minimal surfaces

Isoperimetric Inequalities for the Cauchy-Dirichlet Heat Operator

OPTIMAL POTENTIALS FOR SCHRÖDINGER OPERATORS. 1. Introduction In this paper we consider optimization problems of the form. min F (V ) : V V, (1.

Shape optimisation for the Robin Laplacian. an overview

A Brunn Minkowski inequality for the Hessian eigenvalue in three-dimensional convex domain

Boundary value problems for the infinity Laplacian. regularity and geometric results

Probabilistic Methods in Asymptotic Geometric Analysis.

Affine surface area and convex bodies of elliptic type

SPECTRAL PROPERTIES OF THE LAPLACIAN ON BOUNDED DOMAINS

THE STEINER REARRANGEMENT IN ANY CODIMENSION

RANDOM FIELDS AND GEOMETRY. Robert Adler and Jonathan Taylor

A glimpse into convex geometry. A glimpse into convex geometry

Measuring Ellipsoids 1

ON THE HONEYCOMB CONJECTURE FOR ROBIN LAPLACIAN EIGENVALUES

Some isoperimetric inequalities with application to the Stekloff problem

Moment Measures. Bo az Klartag. Tel Aviv University. Talk at the asymptotic geometric analysis seminar. Tel Aviv, May 2013

arxiv: v1 [math.ap] 23 Apr 2018

Estimates in surfaces with positive constant Gauss curvature

A Variational Analysis of a Gauged Nonlinear Schrödinger Equation

Banach Journal of Mathematical Analysis ISSN: (electronic)

THE HOT SPOTS CONJECTURE FOR NEARLY CIRCULAR PLANAR CONVEX DOMAINS

Logarithmic Sobolev Inequalities

The Volume of the Intersection of a Convex Body with its Translates

LECTURE 15: COMPLETENESS AND CONVEXITY

Mass transportation methods in functional inequalities and a new family of sharp constrained Sobolev inequalities

Eigenvalue (mis)behavior on manifolds

Geometric and isoperimetric properties of sets of positive reach in E d

2 (Bonus). Let A X consist of points (x, y) such that either x or y is a rational number. Is A measurable? What is its Lebesgue measure?

Regularity of the optimal set for spectral problems and the vectorial Bernoulli problem

A SHARP STABILITY RESULT FOR THE RELATIVE ISOPERIMETRIC INEQUALITY INSIDE CONVEX CONES. A. Figalli and E. Indrei. Abstract

The location of the hot spot in a grounded convex conductor

Bulletin of the. Iranian Mathematical Society

P(E t, Ω)dt, (2) 4t has an advantage with respect. to the compactly supported mollifiers, i.e., the function W (t)f satisfies a semigroup law:

Spectral Optimization Problems

ALEKSANDROV-TYPE ESTIMATES FOR A PARABOLIC MONGE-AMPÈRE EQUATION

University of Bristol - Explore Bristol Research. Peer reviewed version. Link to published version (if available): 10.

The Saint-Venant inequality for the Laplace operator with Robin boundary conditions

Partial regularity for fully nonlinear PDE

Lecture No 2 Degenerate Diffusion Free boundary problems

Steiner s formula and large deviations theory

Symmetrization of Lévy processes and applications 1

Star bodies with completely symmetric sections

Probability and Measure

Extreme points of compact convex sets

RENORMALIZED SOLUTIONS ON QUASI OPEN SETS WITH NONHOMOGENEOUS BOUNDARY VALUES TONI HUKKANEN

COMPACT STABLE HYPERSURFACES WITH FREE BOUNDARY IN CONVEX SOLID CONES WITH HOMOGENEOUS DENSITIES. 1. Introduction

SHARP BOUNDARY TRACE INEQUALITIES. 1. Introduction

1 Lesson 1: Brunn Minkowski Inequality

NEW BOUNDS FOR THE PRINCIPAL DIRICHLET EIGENVALUE OF PLANAR REGIONS

AFFINE MAXIMAL HYPERSURFACES. Xu-Jia Wang. Centre for Mathematics and Its Applications The Australian National University

GLOBAL LIPSCHITZ CONTINUITY FOR MINIMA OF DEGENERATE PROBLEMS

REGULARITY OF THE OPTIMAL SETS FOR SOME SPECTRAL FUNCTIONALS

Gaussian Measure of Sections of convex bodies

215 Problem 1. (a) Define the total variation distance µ ν tv for probability distributions µ, ν on a finite set S. Show that

On the p-laplacian and p-fluids

Symmetry of nonnegative solutions of elliptic equations via a result of Serrin

On John type ellipsoids

Γ Convergence of Hausdorff Measures

Convex shape optimization for the least biharmonic Steklov eigenvalue

On a weighted total variation minimization problem

3 (Due ). Let A X consist of points (x, y) such that either x or y is a rational number. Is A measurable? What is its Lebesgue measure?

ROBUSTNESS OF THE GAUSSIAN CONCENTRATION INEQUALITY AND THE BRUNN-MINKOWSKI INEQUALITY

CONNECTIONS BETWEEN A CONJECTURE OF SCHIFFER S AND INCOMPRESSIBLE FLUID MECHANICS

arxiv: v1 [math.fa] 6 Nov 2017

Uniqueness of ground state solutions of non-local equations in R N

Hyperplane sections of convex bodies

Fractional Perimeter and Nonlocal Minimal Surfaces

Math 321 Final Examination April 1995 Notation used in this exam: N. (1) S N (f,x) = f(t)e int dt e inx.

HW1 solutions. 1. α Ef(x) β, where Ef(x) is the expected value of f(x), i.e., Ef(x) = n. i=1 p if(a i ). (The function f : R R is given.

New Bounds for the Principal Dirichlet Eigenvalue of Planar Regions

Brunn Minkowski Theory in Minkowski space-times. François Fillastre Université de Cergy Pontoise France

A LITTLE REAL ANALYSIS AND TOPOLOGY

MATH 202B - Problem Set 5

Combinatorics in Banach space theory Lecture 12

On Mahler s conjecture a symplectic aspect

Transcription:

Shape optimization problems for variational functionals under geometric constraints Ilaria Fragalà 2 nd Italian-Japanese Workshop Cortona, June 20-24, 2011

The variational functionals The first Dirichlet eigenvalue of the Laplacian { Ω λ 1 (Ω) := inf u 2 dx Ω u2 dx : u H 1 0 (Ω), The torsional rigidity 1 { τ(ω) := inf Ω u 2 dx ( ) 2 : u H0 1 (Ω), Ω u dx Ω Ω } u 2 dx > 0 } u dx > 0 The Newtonian capacity (for n 3) { } Cap(Ω) := inf u 2 dx : u C0 (Rn ), u χ Ω R n \Ω Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 2 / 28

For any of these functionals F it holds: F is a Dirichlet energy, F(Ω) = u Ω 2 dx { u = λ1 (Ω)u in Ω u = 0 { u = 1 in Ω u = 0 u = 0 in R n \ Ω u = 1 u(x) 0 as x + Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 3 / 28

F is monotone by inclusions F is homogeneous under dilations (of degree α = 2, n + 2, n 2) F is continuous with respect to Hausdorff convergence F is shape differentiable: if Ω t = (I + tv )(Ω), d dt F(Ω t) t=0 = ± (V ν) u Ω 2 dh n 1 Ω Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 4 / 28

The geometric constraints The volume Ω The perimeter Ω (for sets with finite perimeter, χ Ω BV ) The mean width (for convex sets, Ω = int(k )) w K (ξ) := h K (ξ) + h K ( ξ), h K (ξ) := sup x K (x ξ) for ξ S n 1 the distance between the two support planes of K normal to ξ M(K ) := 2 H n 1 (S n 1 ) S n 1 h K (ξ) dh n 1 (ξ). Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 5 / 28

The problems under study Find extremal domains for F(Ω) = λ 1 (Ω), τ(ω), Cap(Ω) under one of the constraints Ω, Ω, M(Ω) = const. Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 6 / 28

The meaningful problems are: min λ 1(Ω) Ω =c max τ(ω) Ω =c min Cap(Ω) Ω =c min λ 1(Ω) Ω =c max τ(ω) Ω =c min Cap(Ω) Ω =c min λ 1(Ω) M(Ω)=c max τ(ω) M(Ω)=c max Cap(Ω) M(Ω)=c We are interested as well in finding stationary domains for these problems. Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 7 / 28

Outline of the talk 1. Volume constrained problems 2. Perimeter constrained problems 3. Mean width constrained problems 4. Some results about a conjecture by Pólya-Szegö Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 8 / 28

1. Volume constrained problems Assume Ω = B. Then: λ 1 (Ω) λ 1 (B) [Faber-Krahn] τ(ω) τ(b) [Pólya] Cap(Ω) Cap(B) [Szegö] Proof. By Schwarz symmetrization. Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 9 / 28

Under the assumpti C 2, if there exists a solution C 2 up to the boundary to any of the following overdetermined b.v.p., necessarily Ω = B: u = 1 u = 0 u = c ν in Ω u = λ 1 (Ω)u u = 0 u = c ν in Ω [Serrin 71] [Reichel 97] u = 0 in R n \ Ω u = 1 u 0 as x + u = c ν Proof. By moving planes or by many different methods! Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 10 / 28

2. Perimeter constrained problems Assume Ω = B. Then the extremality of balls under volume constraint, combined with the isoperimetric inequality yields: λ 1 (Ω) λ 1 (B) τ(ω) τ(b) Ω 1/n Ω 1/(n 1) B 1/n B 1/(n 1), Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 11 / 28

Under the assumpti C 2, if there exists a solution C 2 up to the boundary to any of the following overdetermined b.v.p., necessarily Ω = B: u = 1 u = 0 u 2 = c H Ω ν [Serrin 71] in Ω u = λ 1 (Ω)u u = 0 u 2 = c H Ω ν in Ω Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 12 / 28

Conjecture: [Pólya-Szegö 51] Among convex bodies K R 3, with given surface measure S(K ), the planar disk D minimizes the Newtonian capacity. The convexity constraint is irremissible! S(K ) it is meant as H 2 ( K ) if int(k ) and 2H 2 (K ) otherwise. The solution cannot be the ball!! Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 13 / 28

Conjecture: [Crasta-F.-Gazzola 05] Among open smooth and strictly convex sets, balls are the unique stationary domains for the PS problem. u = 0 in R n \ Ω u = 1 u 0 as x + u 2 = c H Ω ν Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 14 / 28

3. Mean width constrained problems Remark: The mean width is a Minkowski linear functional. Recall that K + L is defined by the equality h K +L = h K + h L. Concavity inequalities in the Minkowski structure: F(Ω) = λ 1 (Ω), τ(ω), Cap(Ω) satisfy a Brunn-Minkowski type inequality: F 1/α (K + L) F 1/α (K ) + F 1/α (L) K, L K n, with strict inequality for non-homothetic sets. [Brascamp-Lieb 73, Borell 83, 85, Caffarelli-Jerison-Lieb 96, Colesanti 96] Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 15 / 28

Theorem (shapeopt under mean width constraint). [Bucur-F.-Lamboley 11] Assume that F : K n R + satisfies a BM-type inequality, is invariant under rigid motions, and continuous in the Hausdorff distance. Consider the quotient E(K ) := F 1/α (K ) M(K ). Then: (i) the maximum of E over K n is attained only on balls; (ii) if n = 2, the minimum of E over K 2 can be attained only on triangles or on segments. In particular, if M(Ω) = M(B): λ 1 (Ω) λ 1 (B) τ(ω) τ(b) Cap(Ω) Cap(B) Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 16 / 28

Proof (i) Hadwiger s Theorem: For every K K n (with dim K > 0 ) there exists a sequence K h of rotation means of K which converges in Hausdorff distance to a ball. Then: F 1/α (K ) M(K ) F 1/α (K h ) M(K h ) F 1/α (B) M(B) (ii) In dimension n = 2, the unique Minkowski indecomposable bodies are triangles and segments. (No longer true in higher dimensions!) Then: if K is not a triangle or a segment, K = K 1 + K 2 = F 1/α (K ) M(K ) > F 1/α (K 1 ) + F 1/α (K 2 ) M(K 1 ) + M(K 2 ) { F 1/α (K 1 ) min M(K 1 ), F 1/α (K 2 ) }. M(K 2 ) Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 17 / 28

Theorem (Gaussian curvature overdetermined b.v.p.) [F. 11] Under the assumption Ω = int K for some K K n 0, with Ω of class C2, if there exists a solution C 2 up to the boundary to any of the following overdetermined b.v.p., necessarily Ω = B: u = 1 u = 0 u 2 = c G Ω ν in Ω u = λ 1 (Ω)u u = 0 u 2 = c G Ω ν u = 0 in R n \ Ω u = 1 u 0 as x + u 2 = c G Ω ν in Ω Proof. By concavity, a stationary domain for the quotient functional E = F 1/α M is necessarily a maximizer. Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 18 / 28

Concavity inequalities in the Blaschke structure: We say that F : K0 n R+ satisfies a Kneser-Süss type inequality if F (n 1)/α (K L) F (n 1)/α (K ) + F (n 1)/α (L) K, L K0 n, with equality if and only if K and L are homothetic. K L is defined by the equality σ(k L) = σ(k ) + σ(l), where σ(k ) := (ν K ) (H n 1 K ) ν K = Gauss map Kneser-Süss Theorem states that the above concavity inequality holds true for the volume functional. S(K ) is a Blaschke linear functional Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 19 / 28

Theorem (shapeopt under surface contraint). [Bucur-F.-Lamboley 11] Assume that F : K0 n R+ satisfies a KS-type inequality, is invariant under rigid motions and continuous in the Hausdorff distance. Consider the quotient E(K ) := F (n 1)/α (K ). Then: S(K ) (i) the maximum of E over K0 n is attained only on balls; (ii) the minimum of E over K0 n can be attained only on simplexes. Counterexamples: F(Ω) = Cap(Ω), λ 1 (Ω), τ(ω) do not satisfy a KS-type inequality! Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 20 / 28

Concavity inequalities in different algebraic structures: each of our model functional is concave with respect to a new sum of convex bodies, which linearizes the first variation of F. This leads to new isoperimetric-like inequalities. Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 21 / 28

4. Some results about PS conjecture inf E(K ) := Cap2 (K ) K K 3 S(K ) Theorem 1 (optimality of the disk among planar domains). [Pólya-Szegö, 51] Let D be a planar disk. For every planar convex domain with H 2 (K ) = H 2 (D), it holds Cap(K ) Cap(D). Proof. By a cylindrical symmetrization. Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 22 / 28

Theorem 2 (lower bound). [Pólya-Szegö, 51] The infimum of E over K 3 is strictly positive. Proof. By using symmetrizations and monotonicity with respect to inclusions. Theorem 3 (existence of a minimizer). [Crasta-F.-Gazzola, 05] The infimum of E over K 3 is attained. Proof. By using Blaschke selection theorem, John Lemma, and the behaviour of thinning ellipsoids. Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 23 / 28

Theorem 4 (optimality among ellipsoids). [F.-Gazzola-Pierre 11] The planar disk is optimal for E within the class of triaxial ellipsoids. Proof. Plot of the map (b, c) E 1 (E 1,b,c ) for (b, c) in the triangle T = {(b, c) R 2 : 1 b c 0} and for (b, c) near (1, 0). c 0.5 0.0 c 0.5 0.0 1.0 1.0 1.0 0.5 b Remarks: (i) There is no stationary ellipsoid different from a ball. (ii) There exists b s.t. c E(E(1, b, c)) is not monotone. 0.0 Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 24 / 28 1.0 0.8 b 0.6 0.4

Theorem 5 ( local optimality). [F.-Gazzola-Pierre 11] For a large class of suitably defined one parameter families of convex domains D t obtained by fattening the planar disk, it holds E(D) < E(D t ) for 0 < t 1. Proof. By a careful comparison of Cap (0) and S (0). Left: Cap (0) > 0, S (0) = 0. Right: Cap (0) = +, S (0) < +. Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 25 / 28

Theorem 6 (no smooth portions with positive Gauss curvature). [Bucur-F.-Lamboley 11] Assume that F : K n 0 R+ is given by F(K ) = f ( K, λ 1 (K ), τ(k ), Cap(K )) ( with f C 2 ). Let K be a minimizer over K0 n for the functional E(K ) := F(K ) S(K ). If K contains a subset ω of class C 3, then G K = 0 on ω. Proof. l S 2 (K ) (ϕ, ϕ) c 1 ϕ 2 H 1 (ω) +c 2 ϕ 2 l L 2 (ω), F 2 (K ) (ϕ, ϕ) c 3 ϕ 2. H 1 2 (ω) Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 26 / 28

Lemma (local concavity entails local extremality). Let K K n 0 be a minimizer for J : Kn 0 R+. Let ω K of class C 3, and assume that t J(K t ) is twice differentiable at t = 0 for any smooth V compactly supported on ω. If the bilinear form l J 2 (K ) satisfies: ϕ C c (ω), l J 2 (K ) (ϕ, ϕ) c 1 ϕ 2 H 1 (ω) + c 2 ϕ 2 H 1 2 (ω) for some constants c 1 > 0, c 2 R, then G K = 0 on ω Proof. By contradiction, against the second order optimality condition. Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 27 / 28

The end. Thank you! Ilaria Fragalà (Politecnico di Milano) Cortona, June 23, 2011 28 / 28